Fractal-Apps

Replace device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu") by device = torch.device("cpu")

Thanks. For information, I've installed scann manually from a wheel file, and it works in Windows.

You can download it from here: https://github.com/google-research/google-research/tree/master/scann I've discovered another solution since then: install ubuntu on windows: https://ubuntu.com/tutorials/install-ubuntu-on-wsl2-on-windows-10#1-overview It also works with Docker on Windows as it is linux based.

I don't remember exactly how I've installed it. All I remember is that I've downloaded the full directory using this service: https://download-directory.github.io/ And then install manually applying the paragraph "building...

I have the same issue because of RAM, not because of GPU.

Same issue. Python 3.8.5. It seems to be because of windows 11. Scann is a Linux library.

I've found a solution to install it manually. See: https://github.com/CompVis/latent-diffusion/issues/147#issuecomment-1279934409

No problem indeed if the images have the same size. But what if want to use the model for many images having different sizes? I don't want to load the...

Yes, I've thought about it: it is a very good workaround but it cannot cover all cases in a clean way and it uses additional memory :-/

You should be able to sign with dkimpy, but the documentation is poor.